An Improved Multi-Population Immune Genetic Algorithm

To overcome the shortcomings of traditional genetic algorithms (GAs), a novel multi-population immune genetic algorithm (MPIGA) is proposed, which introduces some mechanisms of immune system into GA, including antigen recognition, immune memory and concentration regulation, and an elite inheritance strategy of antibody in memory cells is also used to ensure the convergence of MPIGA. At the same time, based on the theory of multi-population evolution, MPIGA separates antibody competition into two steps, competition among sub populations and competition among individuals in a sub population, which can resolve the conflict between global and local searching abilities. Experimental results of optimizing some typical test functions demonstrate that the MPIGA has superior performances and can converge to the global optimal point more rapidly and stably than other GAs.